[rrg] Summary of architectural solution space

[Brian E Carpenter]

On 2008-11-25 07:07, Noel Chiappa wrote:
>     > From: "Tony Li" <tony.li at tony.li>
> 
>     >>> I thought the standard model for IPv6 was to assign sites a PI /48?
> 
>     >> No, that's the standard heresy. The standard model has always been
>     >> multiple PA's.
> 
>     > It seems that the IAB then needs to have a chat with the RIR's then.
> 
> I think that horse escaped rather a long time ago, alas.

Indeed. Actually it escaped via RFC 1881 (December 1995), and it was
intentional. I personally believe that the registries have made it
too easy to get PI prefixes, but that was a choice of the operational
community. However, it doesn't change what I was trying to say,
which is that the *design* assumption for IPv6 was PA, so it seems
appropriate to call that Plan A.

   Brian

> If the RIR's were
> willing to think they were able to make routing engineering decisions, I
> doubt very much they'd be willing to reverse course because the IAB told them
> to.
> 
> The irony here is that the adoption of PI was seen as 'necessary' to drive
> the adoption of IPv6. As some of us pointed out at the time, it would in fact
> likely have the _opposite_ impact, by making the routing overhead of IPv6
> even higher for the ISPs.
> 
> I think the old Benjamin Franklin quote about experience applies here as to
> so many other aspects of IPv6.
> 
> 	Noel
> _______________________________________________
> rrg mailing list
> rrg at irtf.org
> https://www.irtf.org/mailman/listinfo/rrg
>